Mapping the AI Money Web: Inside the $Trillion Circle of Chips, Clouds, and Capital

If you've tried to trace who actually owns, funds, or supplies whom in the AI industry lately, you've probably given up somewhere around the fifth overlapping deal. That's not an accident — it's the defining structural feature of this AI cycle. A handful of companies are simultaneously investors, suppliers, and customers of one another, and the resulting web now touches nearly every trillion-dollar company on the planet. Here's how the money actually moves.

The center of gravity: Nvidia

At roughly $4.8 trillion in market value, Nvidia isn't just the industry's chip supplier — it's become its largest venture investor. Its equity portfolio includes a $30 billion stake in OpenAI, up to $10 billion in Anthropic, and up to $2 billion in xAI, alongside smaller bets on infrastructure players like CoreWeave, Nscale, Nebius, and Figure AI. Bloomberg has estimated Nvidia's total proposed AI partnerships and financing arrangements — including debt guarantees — at over $750 billion, though only a portion represents capital genuinely at risk.

Notably, Nvidia's CEO signaled in March 2026 that the company's roughly $40 billion in combined OpenAI and Anthropic bets may be its last major private checks into either lab, framing it as a natural close of the investment window ahead of expected 2026 IPOs — not a change of heart on the sector.

OpenAI: the $852 billion hub

OpenAI's most recent funding round valued the company at $852 billion, built on a $110 billion raise in February 2026 — $30 billion from Nvidia, $30 billion from SoftBank, and $50 billion from Amazon. Microsoft remains its earliest and largest strategic backer, holding roughly 49% of the for-profit subsidiary via cumulative investment north of $13 billion.

On the compute side, OpenAI has stacked multiple mega-deals simultaneously: Oracle: reportedly a ~$300 billion cloud-compute agreement over five years, part of the broader $500 billion Stargate data center project with SoftBank Nvidia: a letter of intent for at least 10 gigawatts of systems, with investment tied progressively to deployment AMD: a deal to deploy up to 6 gigawatts of AMD Instinct GPUs, sweetened by a warrant for up to 160 million AMD shares — potentially a ~10% stake in AMD if fully exercised

Anthropic: the fastest-rising node

Once valued well below OpenAI, Anthropic's May 2026 Series H raised $65 billion at a $965 billion post-money valuation, putting it ahead of OpenAI's private mark on a valuation basis (the $380B figure in circulation reflects an earlier round). Its ownership map reads like a hyperscaler consortium: Amazon: ~$13 billion invested, with up to $20 billion more pledged on commercial milestones Google: a verified 14% stake, capped at 15%, alongside up to $40 billion in further planned investment Microsoft: a $5 billion investment made in November 2025 as part of a deal under which Anthropic committed to $30 billion in Azure spend

Collectively, Amazon, Google, and Microsoft are estimated to hold 34–45% of Anthropic — while Anthropic has committed to spend over $75 billion buying cloud infrastructure back from those same three companies.

Why this matters: the "circular deal" problem

The pattern across nearly every one of these relationships is the same: a hyperscaler or chipmaker invests equity in an AI lab, and the lab commits to spend a comparable — often larger — sum buying compute back from the investor. Nvidia funds OpenAI, which buys Nvidia chips. Microsoft and Amazon fund Anthropic, which buys Azure and AWS capacity. It's a structure that props up reported earnings (Microsoft, Amazon, and Google have all booked billions in paper gains from these stakes) while raising a genuine question analysts have started asking openly: how much of this is real demand, versus capital recycling through a closed loop?

The IMF flagged exactly this concern in its July 2026 World Economic Outlook Update, warning of froth in AI-exporting economies. Nvidia's own credit default swap spreads saw their sharpest single-day move on record in July 2026, after reports the company was weighing a financing guarantee tied to OpenAI's chip purchases — a sign that debt markets, at least, are starting to price in circularity risk.

The takeaway for investors and operators

Concentration risk is real — a small cluster of firms (Nvidia, Microsoft, Google, Amazon, OpenAI, Anthropic) now accounts for a disproportionate share of both AI capex and AI-linked corporate earnings growth. One CNBC analysis found that stripping out investment gains from Anthropic and OpenAI stakes cut aggregate S&P 500 earnings growth from ~48% to ~29% in a recent quarter.

Diversified compute strategies are becoming standard practice — OpenAI's simultaneous deals with Nvidia, AMD, and Oracle, and Anthropic's presence across AWS, Azure, and Google Cloud, both reflect a deliberate hedge against single-supplier dependency.

Watch the IPO calendar — both OpenAI and Anthropic are widely expected to test public markets in 2026–27. That will be the first moment these valuations and cross-holdings face independent market pricing rather than private negotiation.

Equity-for-compute structures (like AMD's OpenAI warrant) are becoming a template — expect more chipmakers and infrastructure players to use equity sweeteners to lock in anchor customers, rather than pure cash sales.

The AI buildout is real, and so is the demand behind it. But the closer you look at who's actually funding whom, the more this looks less like a straightforward customer market and more like a small number of players financing each other's growth — with the rest of the market watching to see whether the underlying revenue eventually catches up to the capital.

Sources: CNBC, Axios, Reuters, TechCrunch, Yahoo Finance/Fortune, AMD & OpenAI press releases